The Future. As deepfakes become more ubiquitous online, they could force the music industry to consider whether artists own the sounds produced by their vocal cords. Instead of punishing copycats, labels might gain more from embracing the new tech and finding ways to monetize it.
What’s the new tech anyway?
So-Vits-SVC is a free, open-source voice model that can run locally on any computer. Since launching in March, the AI has gotten easier to use as updated versions arrive almost daily, per Vulture.
- A good model can perform any song as long as the user has isolated vocal and instrumental tracks. If not, other programs can help separate them.
- For best results, the AI clone should have the same vocal range as the artist it’s simulating. It’ll follow their inflections, so singers with accents or other distinctive traits might be harder to replicate.
How can musicians take advantage?
While no software can write good music and lyrics on its own, the best AI-generated songs still require the human touch. But together, the possibilities are intriguing.
- Many artists already outsource their songwriting and could become wealthier by lending their voices to new material — including the estates of dead artists who can no longer release new music.
- Talented unknowns could finally make money from their music.
- AI-generated vocals could provide backup for artists whose voices have changed over time.